The broadband networks have taken shape currently. FIG. 1 is a schematic diagram illustrating the network architecture of a broadband network in the prior art: the broadband network includes an RG 11, a Digital Subscriber Line Access Multiplexer (DSLAM) 12, a BRAS13, a core network and a network management unit 14, which are interconnected successively. The broadband network may be divided into three domains: a Digital Subscriber Line (DSL) domain between the RG11 and the DSLAM12; an Ethernet domain between the DSLAM12 and BRAS13; a management domain between the BRAS13 and the network management unit 14. There may be an Ethernet Switch performing the service switching function in the Ethernet domain, i.e., between the DSLAM12 and the BRAS13.
Herein, the DSLAM12 is a device with perfect asynchronous transmission mode (ATM) architecture, i.e. the ATM-based DSLAM. As the existing metropolitan area ATM core network is not extended any more currently, while the Internet Protocol (IP) broadband metropolitan area core network is new founded, the ATM-based DSLAM will be more and more difficult to suit the requirement of actual networking, along with the available resources of metropolitan area ATM core network decrease day by day. As a result, the DSLAM12 has gradually been evolving from an ATM-based DSLAM into an IP-based DSLAM, thereby enabling the use of a broadband IP network which can provide services, such as, Broadcast Television (BTV), Video On Demand (VOD), voice, high-speed internet access, Virtual Private Network (VPN). Along with the development of various services in broadband IP networks, the management of Service Level Agreement (SLA) will be desired, that is, the serving with different levels are provided to the service with different levels in broadband IP networks.
One main aspect of the performance management between the RG11 and the BRAS13 is to perform SLA tests. In addition, another aspect of the performance management between the RG11 and the BRAS13 is to collect the statistical parameters information of a traffic flow and provide it to an Operation Support System (OSS) for performance analysis and accounting.
In a broadband IP network with an IP-based DSLAM, Ethernet protocols are used in the Ethernet domain between a DSLAM and a BRAS, and ATM or point-to-point Ethernet protocols are used in the DSL domain between an RG and a DSLAM. If the end-to-end performance management is performed between an RG and a BRAS, which requires the RG to support the service layer Ethernet Operation Administration and Maintenance (OAM) protocol. It is difficult to support the service layer Ethernet OAM protocol for the RG in the initial creating phase of a broadband IP network, at present, the service layer Ethernet OAM protocol is only implemented on the DSLAM and the BRAS.
In order to implement the performance management between an RG and a BRAS, currently the statistical parameters information of a traffic flow is mainly extracted by a network management, namely, the network management enquires the statistical parameters information of a traffic flow of every device, for example, the RG, the DSLAM and the BRAS; in order to implement the SLA test, the indexes can be fragmentally measured, such as delay, jittering or packet-loss.
However, the network management performs the flow statistics to every device respectively, and the results are inaccurate, for example, in order to perform the statistics of the traffic flow between an RG and a BRAS, and the statistics of the traffic flow is respectively performed to the RG, to the DSLAM and to the BRAS. The statistical result of the traffic flow in each device can not represent the real flow, which leads to an inaccurate result in the statistics of the traffic flow. And the statistics of the traffic flow need to be manually performed in every device, which is very boring and trivial; in the SLA test, since the SLA test is fragmentally performed, the indexes measured fragmentally, i.e. delay, jittering or packet-loss, can not form the exact end-to-end SLA indexes between the RG and the BRAS, and the results measured fragmentally can not really reflect the situation of end-to-end service operation between the RG and the BRAS, which leads to an inaccurate result in the SLA test, and performing the SLA test fragmentally is very boring and trivial.